Quality and Safety Improvement in Spine Surgery

Multimodal machine learning for predicting perioperative safety indicators in spinal surgery

BACKGROUND CONTEXT

Machine learning (ML) algorithms can utilize the large amount of tabular data in electronic health records (EHRs) to predict perioperative safety indicators. Integrating unstructured free-text inputs via natural language processing (NLP) may further enhance predictive accuracy.

PURPOSE

To design and validate a preoperative multimodal ML architecture that integrates structured EHR data (patient demographics, comorbidities, and clinical covariates) with unstructured free-text inputs (past medical and surgical history, medications, and problem lists) via NLP. The multimodal models aim to improve the prediction of perioperative safety indicators compared to baseline ML models that only use structured tabular EHR data.

STUDY DESIGN

Retrospective cohort study.

PATIENT SAMPLE

1,898 patients admitted for elective or emergency spine surgery at four separate large urban academic spine centers during a 5-year period from 2018 to 2023.

OUTCOME MEASURES

Numerical outputs between 0 and 1 corresponding to the likelihood of (I) extended length of stay (LOS), (II) 90-day reoperation, and (III) perioperative intensive care unit (ICU) admission.

METHODS

We predicted the following safety indicators (I) extended length of stay (LOS), (II) 90-day reoperation, and (III) perioperative intensive care unit (ICU) admission. The quanteda package for NLP within the R environment was utilized to preprocess free-text EHR inputs. The refined text was tokenized and transformed into numerical vectors using a bag-of-words approach and integrated with the tabular EHR data to create a document-feature matrix. Two extreme gradient boosted (XGBoost) ML models were trained: a base model utilizing only structured tabular EHR data and a combined multimodal model that leveraged both combined structured tabular EHR data with numerical vectors derived from free-text NLP inputs. Hyperparameter tuning was performed via grid search, and the models were validated using 10-fold cross validation with an 80:20 training/testing split. Word clouds were generated for the free-text data and explainable artificial intelligence (XAI) techniques were employed for feature importance. Metrics calculated for model performance included Area Under the Receiving-Operating Characteristic Curve (AUC-ROC), Brier score, Calibration slope, Calibration Intercept, Precision, Recall and F1-Score.

RESULTS

1,898 patients (60.7% female) were extracted from January 2018 to September 2023, with a median age of 60.0 (IQR: 52.0-68.0) and median body mass index (BMI) of 30.3 kgm2 (IQR: 26.3-34.6). Extended LOS was defined as ≥ 14.4 days, constituting 10.1% of all individuals. The median LOS for the entire cohort was 4.0 days (IQR: 2.0-7.0), while the 90-day reoperation rate was 10.54%, and the ICU admission rate was 7.74%. The preoperative tabular EHR models predicted perioperative safety indicators with AUC ranging from 0.770 to 0.779, Brier scores ranging from 0.074 to 0.099, and calibration slopes ranging from 2.279 to 2.418. Precision and recall for this model ranged from 0.918 to 0.973 and 0.988 to 0.994, respectively, resulting in F1-scores between 0.954 and 0.973. The combined multimodal models predicted perioperative safety indicators with AUC ranging from 0.827 to 0.903, Brier scores ranging from 0.056 to 0.083, and calibration slopes ranging from 0.755 to 1.217. The multimodal models achieved precision ranging from 0.909 to 0.933 and recall ranging from 0.979 to 0.994, leading to F1-scores between 0.943 and 0.962. Important tabular predictors included patient age, BMI, hemoglobin level, white blood cell count, platelet count, and a combined anterior/posterior spinal fusion approach. Important free-text inputs included vertebral osteomyelitis, radiculopathy, myelopathy, and spinal metastasis.

CONCLUSIONS

The multimodal NLP model exhibited superior performance in all outcome measures when compared to the baseline tabular model. Future work includes incorporating additional model dimensions, such as the history of present illness, physical exam, and spinal imaging, and clinically implementing the models into our informed consent and preoperative optimization pathway.

The use of machine learning for the prediction of response to follow-up in spine registries

BACKGROUND

One of the main challenges in the maintenance of registries is to keep a high follow-up rate and a reliable strategy to limit dropout is currently lacking. Aim of this study was to utilize machine learning (ML) models to highlight the characteristics of patients who are most likely to drop out, and to evaluate the potential cost effectiveness of the implementation of a follow-up system based on the obtained data.

METHODS

All patients recruited in the local spine surgery registry were included and demographic, peri- and postoperative data were collected. Five ML models were trained and evaluated for response to follow-up prediction. Explainable and Cautious AI were then implemented to increase the trustworthiness of the model. The efficacy and cost effectiveness of the current follow-up strategy (call everybody) were compared to a strategy based on the implemented model (call only patients with high dropout risk).

RESULTS

Records from 4652 patients were available. The random forest (RF) outperformed all models in the prediction of response to follow-up. Among the considered variables, the ones that had the most weight were length of follow up, level of the main pathology and extent of surgery, SF-36 and BMI. Interpretable Decision Trees (IDT) and selective prediction models further increased the performance of the model. The cost reduction calculation predicted that implementing the developed ML model in the clinical practice would, over time, result in a reduction of costs by 31%, with only 2‰ missed calls.

CONCLUSION

ML models can effectively identify patients with high risk of dropout. The RF model outperformed all evaluated models, and was further improved with the use of Controllable AI. The application of ML to the follow-up strategy could reduce costs and limit missed responses.

Exploring strategies to enhance patient safety in spine surgery: a review

Patient safety is the foundation of spine surgery, where the intricate nature of spinal procedures and the unique risks involved call for exceptional diligence and comprehensive protocols. In this high-stakes field, developing and implementing rigorous safety protocols is not only vital for minimizing complications but also for achieving the best possible outcomes and strengthening the confidence patients have in their care team. Each patient entrusts their well-being to their surgical team. This trust underscores the responsibility healthcare providers have to prioritize safety at every stage. In spine surgery, thorough preoperative planning, clear communication during informed consent, and vigilant postoperative care are all crucial for creating a safe environment tailored to each patient's needs. A commitment to patient safety requires more than individual efforts; it calls for a coordinated, multidisciplinary approach where surgeons, nurses, anesthesiologists, and rehabilitation specialists work closely together. This collaboration ensures that each step of the patient's journey is aligned with best practices for safety and care. This review highlights the critical need for ongoing evaluation and refinement of safety protocols in spine surgery. As surgical techniques and technologies advance, and as patients' needs evolve, healthcare teams must remain responsive, cultivating a culture of safety that is both proactive and adaptable. Continuous investment in quality improvement and research is essential to fine-tune these protocols, ensuring they remain both relevant and effective in addressing the unique challenges of spine surgery. Prioritizing comprehensive safety measures goes beyond improving surgical outcomes; it plays a pivotal role in strengthening the trust and confidence patients have in their healthcare providers. By committing to these robust protocols, we reaffirm our dedication to patient-centered care, enhancing not only patient safety and recovery but also fostering a deeper faith in a healthcare system that places patient well-being at the forefront.

Sarcopenia and Postoperative Complications, Cost of Care, and All-Cause Hospital Readmission Following Lumbar Spine Arthrodesis: A Propensity Matched Cohort Study

OBJECTIVE

Sarcopenia, characterized by decreased muscle mass and function, is projected to affect more than 200 million people worldwide by 2060. This study aimed to evaluate the rates of short-term complications following lumbar spine arthrodesis in patients with and without a recent diagnosis of sarcopenia.

METHODS

The PearlDiver database was queried to evaluate all patients who underwent index lumbar spine arthrodesis from 2012 to 2019. Multivariate logistic regression was used to compare rates of 90-day surgical and medical complications. Kaplan-Meier analysis was performed to compare cumulative rates of reoperation and all-cause hospital readmission. Two sample t testing was used to compare costs of care. Statistical significance was set at P < 0.05 a priori.

RESULTS

Of 239,953 patients undergoing lumbar spine arthrodesis, 1087 had a recent diagnosis of sarcopenia (0.45%) before surgery. Patients with sarcopenia were significantly more likely to experience a urinary tract infection (odds ratio = 1.41, P = 0.035) and undergo incision and drainage (odds ratio = 2.66, P = 0.010) within 90 days after lumbar arthrodesis. Patients with sarcopeniawere at a 24% greater risk of 1-year all-cause hospital readmission. The 90-day cost of care was significantly greater in patients with sarcopenia ($37,689.86 vs. $26,635.72; P < 0.001).

CONCLUSIONS

In patients undergoing lumbar spine arthrodesis, sarcopenia is associated with an increased risk of postoperative complications, including increased costs of care. Spine surgeons should consider screening patients for sarcopenia preoperatively and counsel them regarding their increased risk of complications. Additionally, surgeons may consider preoperative optimization, like the management of low bone density.

Single-Level Total Disc Replacement: Index-Level and Adjacent-Level Revision Surgery Incidence, Characteristics, and Outcomes

BACKGROUND

The literature reports that index level (IL) revision spine surgery (RSS) and adjacent level (AL) RSS are diminished in lumbar TDR compared with fusion procedures. There is a paucity of PROMs reported after RSS.

OBJECTIVE

To present the incidence of RSS at the IL and AL following single-level lumbar total disc replacement (TDR) and to document patient-related outcome measures (PROMs) associated with RSS.

METHODS

PROMs and timelines were analyzed for 32 RSS patients from a prospective cohort study of 401 patients treated with TDR for single-level degenerative disc disease. The data collected prospectively are analyzed from baseline (prior to index surgery) to latest follow-up following RSS. PROMs, including visual analog scales for back and leg, Oswestry Disability Index, and Roland-Morris Disability Questionnaire, were collected preoperatively; postoperatively at 3, 6, and 12 months; and annually thereafter until RSS. The time to RSS was recorded, and PROMs for RSS (IL, AL, or both) were documented, analyzed, and compared.

RESULTS

The median time to RSS in the IL cohort was 35 months (interquartile range [IQR] = 9-51 months). The median time to RSS cohort was 70 months (IQR = 41.3-105.3 months). Timepoints facilitate PROM discussion for RSS. Patients in both groups achieved thresholds for the minimum clinically important difference for pain and disability scores. The small sample size in each group contributed to the variability demonstrated by the 95% CIs, thereby cautioning definitive conclusions.

CONCLUSIONS

This study reveals that statistically significant and modest clinical improvements in PROMs can be achieved in RSS for lumbar TDR at IL and AL. The surgical approach and technique are reflective of the pathology and suggest that anterior RSS for AL degeneration and posterior RSS for IL pathology yield similar results.

CLINICAL RELEVANCE

Statistical and clinical improvements can be achieved in IL-RSS and AL-RSS following single level TDR. It is essential for clinicians to understand and verify the underlying IL and/or AL pathology to select an appropriate management strategy and to facilitate balanced informed discussions with patients.

LEVEL OF EVIDENCE: 4

The impact of case order and intraoperative staff changes on spine surgical efficiency

BACKGROUND CONTEXT

Despite concerted efforts toward quality improvement in high-volume spine surgery, there remains concern that increases in case load may compromise the efficient and safe delivery of surgical care. There is a paucity of evidence to describe the effects of spine case order and operating room (OR) team structure on measures of intraoperative timing and OR efficiency.

PURPOSE

This study aims to determine if intraoperative staff changes and surgical case order independently predict extensions in intraoperative timing after spinal surgery for spondylotic diseases.

STUDY DESIGN/ SETTING

Retrospective cohort analysis PATIENT SAMPLE: All patients over age 18 who underwent primary or revision decompression and/or fusion for degenerative spinal diseases between 2017 to 2019 at a single academic institution were retrospectively identified. Exclusion criteria included absence of descriptive data and intraoperative timing parameters as well as surgery for traumatic injury, infection, and malignancy.

OUTCOME MEASURES

Intraoperative timing metrics including total theater time, wheels in to induction, induction start to cut, cut to close, and close to wheels out. Postoperative outcomes included length of hospital stay and 90-day hospital readmissions.

METHODS

Surgical case order and intraoperative changes in staff (circulator and surgical scrub nurse or technician) were determined. Patient demographics, surgical factors, intraoperative timing and postoperative outcomes were recorded. Extensions in each operative stage were determined as a ratio of the actual duration of the parameter divided by the predicted duration of the parameter. Univariate and multivariate analyses were performed to compare outcomes within case order and staff change groups.

RESULTS

A total of 1,108 patients met the inclusion criteria. First, second, and third start cases differed significantly in intraoperative extensions of total theater time, wheels in to induction, induction start to cut, cut to close, and close to wheels out. On regression, decreasing case order predicted extension in wheels in to induction time. Surgeries with intraoperative staff changes were associated with increases in total theater time, induction start to cut time, cut to close time, close to wheels out time, and length of hospital stay. Switch in primary circulator predicted extended theater time and cut to close time. Relief of primary circulator or scrub predicted extended total theater time, induction start to cut time, cut to close time, and close to wheels out time.

CONCLUSIONS

Intraoperative staff change in spine surgery independently predicted extended operative duration. However, higher case order was not significantly associated with procedural time.

The language of all medical publications and spine publications from 1950 to 2020

Background

Excellent research in all fields, including spine surgery, exists in many different regions and languages. This study seeks to determine the relative number of spine related peer-reviewed publications throughout the world based on language.

Methods

Peer-reviewed publications from the eleven most prolific languages in regard to both the number of peer-reviewed spine publications indexed in PubMed and total peer-reviewed publications from 1950-2020 were identified in PubMed.

Results

29,711,547 peer-reviewed publications were analyzed for the languages of interest with 870,404 (3.0%) of those being spine related peer-reviewed publications. Between 1988 and 2019, non-English language peer-reviewed publications decreased annually for both all peer-reviewed publications and spine related peer-reviewed publications by 44% and 36%, respectively. All medical and spine specific peer reviewed publications in English compared to non-English publications have increased by 7.22 and 6.35 times since 1988, respectively. While the ratio of non-English to English spine related publications decreased in all eleven countries, the percentage of the number of spine specific publications written in Chinese (462%), Portuguese (378%), and Spanish (88%) have increased by the listed percentages.

Conclusion

While the proportion of peer-reviewed publications in the field of spine surgery written in English have increased over the past several decades, there are many non-English language peer-reviewed publications each year, particularly in Chinese. Although the rapid increase in the proportion of English spine related publications is beneficial to English speaking physicians and researchers, further research is necessary to understand the impact on non-English speaking physicians and researchers.

Surgical Site Infection Management following Spinal Instrumentation Surgery: Implant Removal vs. Implant Retention: an Updated Systematical Review

Background

The number of lumbar spine surgery increased in recent years. Spinal instrumentation surgery was an integral component in the treatment of spinal pathologies, which can cause surgical site infection (SSI). Surgical site infections (SSIs) are the leading cause of mortality and morbidity after spinal instrumentation surgery. The management of SSI was implant retention and removal is still unclear.

Objective

The objective of this literature is to systematically review the implant removal and retention method for SSI management after spinal instrumentation surgery.

Methods

We searched in PubMed and ScienceDirect for cohort and randomized control trial studies in English, published between 2002 and 2022, which had data on patients with spinal instrumentation surgery. The underlying disease, comorbidities, common bacteria, type of infection, the onset of infection, implant removal, and retention percentage and recommendation were analyzed. Bias analysis using Newcastle-Ottawa Quality Assessment.

Results

We included 15 studies with a total sample were 2.584 with an average of age 15 to 66 years old. The most common organism detected were S. Aureus, MRSA, and S. Epidermis. The most common surgical procedure indications were degenerative followed by scoliosis. Implant removal and retention rate were 0-100% and 0-90,32% respectively. Implant removal is more frequently used in patients after spinal instrumentation surgery than the implant retention method.

Conclusion

Implant retention can be performed in case of SSI is < 3 months after surgery. Implant removal is recommended if the incidence of SSI is > 3 months. Empirical antibiotics therapy is necessary to reduce the possibility of implant removal after debridement. Further studies on the effect of implant removal and retention in patients on infection recurrence, pain, and quality of life of patients are needed.

Safety profile of outpatient vs inpatient ACDF: An analysis of 33,807 outpatient ACDFs

INTRODUCTION

Anterior cervical discectomy and fusion (ACDF) is commonly performed to treat symptomatic cervical spondylolysis. Recently, many spine surgeons have begun performing ACDF in the outpatient setting. However, as this is a relatively new trend, many studies are either outdated or have small sample populations. The aim of this study was to evaluate outcomes following elective outpatient ACDF in comparison to those performed in the inpatient setting.

METHODS

Patients in Mariner Claims Database (2011-2017) undergoing outpatient elective ACDF were propensity score matched using age, gender and comorbidity burden. Chronic and peri-operative complications were assigned based on medical claims codes. All outcomes of interest were analyzed using multivariate logistic regression and compared to those undergoing inpatient ACDF. Significance was defined as p < 0.05 and adjusted with Bonferroni correction.

RESULTS

Outpatient surgery had significantly lower risk of dysphagia within 24 h in both single (OR 0.44, p < 0.001) and multilevel ACDF (OR 0.48, p < 0.001). Patients undergoing outpatient procedures also have lower risk of 90-day minor (Single OR 0.64, p < 0.001; Multilevel OR 0.52, p < 0.001) and major (Single OR 0.48, p < 0.001; Multilevel OR 0.57, p < 0.001) medical complications. Outpatient procedures were also associated with decreased hospital resource utilization with a noted lower risk of subsequent hospital readmission (Single OR 0.71, p < 0.001; Multilevel OR 0.60, p < 0.001) and ER visits (Single OR 0.84, p < 0.001; Multilevel OR 0.87, p < 0.001).

CONCLUSION

Outpatient single and multilevel ACDF may be performed safely in properly selected patients. Since there are relatively low rates of readmission and significant complications within the days following outpatient ACDF, many surgeons should consider transitioning carefully selected patients to an outpatient setting.